[Pipet Devel] Re: Web interface

J.W. Bizzaro bizzaro at bc.edu
Thu Jan 21 13:45:12 EST 1999


Rahul Jain wrote:
> 
> I was thinking about the Web interface we are planning to have to TULIP.
> 
> We'll need to plan out our design of the other tools very carefully so
> that we don't have to create messy kludges to get this web interface to
> work.

My initial thought regarding the Web interface was not that we could or should
replicate the GUI loci in a Web browser.  My thought was that Loci can provide
an HTML interface to _some_ of the _analysis_ loci, considering they are
command-line, short-lived, output ASCI text, which will be formatted into XML anyway.

Konrad and I have communicated quite a bit about the limitations of HTML.  In
fact, when I first mentioned XML, Konrad thought I was speaking of putting
everything into a Web browser.  Neither he nor I like the static interface of
HTML browsers, so I will make a point that GUI loci will not look or act like
one but will be very dynamic.

In short, I think the Web interface can be a quick-and-dirty way for people
without the Loci package to access the wealth of analysis loci we may someday amass.

I appreciate that you looked into this.  But I think this plan would be an
enormous task, and probably best left to some heavy duty Java...which we don't
want to use for reasons I've expressed before.

Do you want to take on this Web interface project, as the simpler project I
envisioned?  I think the biggest part of this would be some sort of XML to
HTML conversion, which is actually what XSL is all about.  The problem with
XSL is that browsers don't support it right now, and I don't think there is a
specification for CML or BSML.  And I think another big thing (providing no
XSL is used) would be converting diagrams specified by XML into GIF format. 
You would have to write a program that would generate custom GIF's.  Check out
NCBI's "graphical view" of nucleotide data:

    http://www.ncbi.nlm.nih.gov/cgi-bin/Entrez/framik?gi=3647232&db=Nucleotide

Also, the Web interface implementation will work from one server that I will
have set up.  It will act as a limited version of Loci, and will contact
certain analysis loci around the Internet.

This is no small project either.  It will almost parallel Loci itself and use
many of the same components.  So, we must limit the types of views and
analysis tools that the Web interface can handle.



Jeff
bizzaro at bc.edu



> 
> Many of the GTK widgets are difficult to incorporate into pure HTML pages.
> Using JavaScript may help, but there may still be some difficulties. It
> may even turn out that we'll need to create a Java interface instead.
> 
> We really ought to limit the widgets we use and the way we modify them.
> - Looking at the glade widget palette, the first 3 rows shouldn't be a
>   problem. Lists are fine as long as they are kept simple.
> - Trees... not too good.
> - Columned lists shouldn't be too bad (either a table or a bunch of lists
>   next to each other w/ Javascript to keep their scrolling in sync... Is
>   that possible?)
> - Columned tree... ugh.
> - Rulers, shouldn't really need those.
> - The H and V rules aren't a problem.
> - The scales and scrollbars, on the other hand present a problem w/out
>   JavaScript.
> - Menu bars are fine, but we may need/want to use layers to implement
>   them.
> - Status bar is easy, so is toolbar.
> - Progress bars may need some JavaScript.
> - Arrows are trivial.
> - Image and pixmap should be simple (as long as my assumptions of what
>   they do are right.)
> - Drawing area, probably kind of tough if my assumption of what it does is
>   right.
> - Font selection, not needed.
> - Most of the Containers can be implemented with tables and/or frames.
> - For scrolled window, viewport, and handle box, I think we'll need
>   layers.
> 
> Maybe we can have wrapper classes around the GTK widgets and then have
> them also able to create HTML code.
> 
> What are everyone's thoughts on this? We really need to make sure our UI
> possibilities are strictly specified and translatable to HTML/JavaScript.
> It may be a pain to do that, I think we want to keep away from requiring
> Java, which will be a real pain.
> 
> --
> -> -\-=-=-=-=-=-=-=-=-=-/^\-=-=-=<*><*>=-=-=-/^\-=-=-=-=-=-=-=-=-=-/- <-
> -> -/-=-=-=-=-=-=-=-=-=/ {  Rahul -<>- Jain   } \=-=-=-=-=-=-=-=-=-\- <-
> -> -\- "I never could get the hang of Thursdays." - HHGTTG by DNA -/- <-
> -> -/- http://photino.sid.rice.edu/ -=- mailto:rahul-jain at usa.net -\- <-
> |--|--------|--------------|----|-------------|------|---------|-----|-|
>       Version 10.423.999.211011001.23.20110101.042
>       (c)1996-1998, All rights reserved.
>       Disclaimer available upon request.

-- 
J.W. Bizzaro                  Phone: 617-552-3905
Boston College                mailto:bizzaro at bc.edu
Department of Chemistry       http://www.uml.edu/Dept/Chem/Bizzaro/
--



More information about the Pipet-Devel mailing list